Meta · llm model

Muse Wonder

Muse Wonder is a Meta language model (public preview) released Oct 12, 2026; Meta's newest language model in our dataset is Muse Spark Open. Proprietary, API access only. Meta's most capable hosted model for long-running agent and coding work, launched Oct 12, 2026 in preview on the Meta Model API and in Muse Code. Verified against vendor sources on Oct 26, 2026.

Reported Last verified Oct 26, 2026 · All models · JSON · Markdown

ReleasedOct 12, 2026
Context window1.05M1,048,576 tokens
LicenceProprietaryProprietary (API)

InTextImageVideoAudioPDFOutText

Key facts

Muse Wonder: key facts with sources
Fact	Value	Source
Vendor	Meta	n/a
Released	Oct 12, 2026	source
Status	public preview	n/a
API id	muse-wonder-1.0	source
Context window	1,048,576 tokens (1.05M)	source
Input	text, image, video, audio, pdf	source
Output	text	source
Licence	Proprietary (API) Hosted on the Meta Model API and in Muse Code; no weights.	source
Pricing	$4 in / $20 out per 1M tokens Standard mode only; no Contributor-mode discount for Muse Wonder.	source

Where Muse Wonder sits in our data

Released Oct 12, 2026 (week 42 of 2026), in October 2026. Other models in our data released that month: GPT-6.1 Terra, Kimi K3.1 Pro, Kimi K3.1 Pro Highspeed, Nemotron 3.5 Bolt E4B (4GB), Voxtral 3, Mercury 3.
Of the 60 models in our data with a published context window, Muse Wonder has the number 14 largest (1.05M tokens). Our data is not every model on the market.
All releases by date: AI model releases by month · new AI models this week. Choosing for code: best LLM for coding.

How Muse Wonder differs from Muse Wonder Full

Read from the recorded facts of the two models, not from a head-to-head test. We have not run either.

We have a recorded price for Muse Wonder but none for Muse Wonder Full.
Muse Wonder lists a 1.05M-token context window; we have none recorded for Muse Wonder Full.
Status is public preview, against limited access.
Both were released on Oct 12, 2026.
Facts for the other model: Muse Wonder Full.

What the sources say

These are claims from the linked pages, not our measurements.

Meta calls Muse Wonder and Muse Wonder Full its most capable models to date, aimed at agentic and coding work.

Notes

Meta's docs describe Muse Wonder as "in preview". It is not available in South Korea.

What we tested

Nothing yet. We have not run Muse Wonder ourselves, so every fact on this page is Reported from the sources below. When we do test it, the scope will be stated here and in our test logs. See methodology.

Not verified, left out

We could not confirm these on a vendor page, so we do not state them.

Parameter count and benchmark scores were not captured.
Meta says a full safety report "will be available soon".